Structure-Function of the Tumor Suppressor BRCA1

BRCA1, a multi-domain protein, is mutated in a large percentage of hereditary breast and ovarian cancers. BRCA1 is most often mutated in three domains or regions: the N-terminal RING domain, exons 11-13, and the BRCT domain. The BRCA1 RING domain is responsible for the E3 ubiquitin ligase activity of BRCA1 and mediates interactions between BRCA1 and other proteins. BRCA1 ubiquitinates several proteins with various functions. The BRCA1 BRCT domain binds to phosphoproteins with specific sequences recognized by both BRCA1 and ATM/ATR kinases. Structural studies of the RING and BRCT domains have revealed the molecular basis by which cancer causing mutations impact the functions of BRCA1. While no structural data is available for the amino acids encoded by exons 11-13, multiple binding sites and functional domains exist in this region. Many mutations in exons 11-13 have deleterious effects on the function of these domains. In this mini-review, we examine the structure-function relationships of the BRCA1 protein and the relevance to cancer progression.

a BRCA2 mutation have a 50-60% lifetime risk of developing breast cancer and a 30% risk of developing ovarian cancer [2]. These genes belong to the tumor suppressor gene family for their capacity to repair damaged DNA through a process known as DNA double-strand break repair [3]. Therefore, an inherited mutation in either of these genes combined with loss of heterozygosity predisposes cells to chromosomal instability and greatly increases the probability of malignant transformation and cancer development. Interestingly, multiple other potential functions have been proposed for the BRCA1 and BRCA2 proteins that may have an impact on their tumor suppressor function [4].
The management of HBOC syndrome is an evolving area, and clearly much more research is needed to understand the molecular basis of cancer progression in these patients. The linkage of BRCA1 and BRCA2 to early-onset hereditary breast cancer was discovered in 1990 and 1994 respectively [5,6]. Since then, BRCA genotyping is now used to determine patient counseling, management decisions, and prognosis of this syndrome [7]. However, inconsistent and limited data exist regarding the clinical course of BRCA-mutated patients after cancer develops [8]. A published meta-analysis for BRCA1related tumors reported a worse outcome among the breast cancer patients carrying a mutated BRCA gene [7], while BRCA1 mutated ovarian cancer patients had a more favorable clinical outcome [9].
Other studies have reported that both BRCA1-mutated breast and ovarian tumors have a better outcome [8,10]. This is likely due to increased sensitivity of BRCA mutated cells to chemotherapeutics targeting DNA such as anti-metabolites, alkylating agents, and topoisomerase inhibitors [11]. However, more research into the molecular basis by which the BRCA proteins function as tumor suppressors and the clinical significance is clearly needed.
Over 1700 unique BRCA1 mutations have been reported to the Breast Cancer Information Core Database [12]. Of these mutations, 858 have been confirmed as being "clinically significant." Clinically significant mutations cause an increased risk of cancer and result in a protein with reduced function or no protein product. Three domains of the BRCA1 protein are mutated in cancer patients with relatively high frequency. These domains include the RING domain (exons 2-7), a region encoded by exons 11-13, and the BRCT domain (exons16-24) (Figure 1). The RING domain functions as an E3 ubiquitin ligase. The amino acids encoded by exons 11-13 contain protein binding domains for a number of diverse proteins. The BRCT domain is a phosphoprotein binding domain with specificity for proteins phosphorylated by ATM/ATR kinases.
Understanding the structural biology of BRCA1 and BRCA2 is important for elucidating both physiologic and pathophysiologic function of these proteins. As shown in Table 1, multiple structures have been solved for the BRCA1 RING and BRCT domains and associated proteins, including clinically relevant mutants.

CSBJ
Abstract: BRCA1, a multi-domain protein, is mutated in a large percentage of hereditary breast and ovarian cancers. BRCA1 is most often mutated in three domains or regions: the N-terminal RING domain, exons 11-13, and the BRCT domain. The BRCA1 RING domain is responsible for the E3 ubiquitin ligase activity of BRCA1 and mediates interactions between BRCA1 and other proteins. BRCA1 ubiquitinates several proteins with various functions. The BRCA1 BRCT domain binds to phosphoproteins with specific sequences recognized by both BRCA1 and ATM/ATR kinases. Structural studies of the RING and BRCT domains have revealed the molecular basis by which cancer causing mutations impact the functions of BRCA1. While no structural data is available for the amino acids encoded by exons 11-13, multiple binding sites and functional domains exist in this region. Many mutations in exons 11-13 have deleterious effects on the function of these domains. In this mini-review, we examine the structure-function relationships of the BRCA1 protein and the relevance to cancer progression.
In this review we will focus on the structural basis by which the BRCA1 protein functions as a tumor suppressor, and highlight the importance of these studies to understanding the pathophysiology and clinical outcomes of breast and ovarian cancers.

RING domain
The RING (Really Interesting New Gene) domain of BRCA1 consists of a RING finger and two flanking alpha helices encompassing amino acids 1-109 (exons 2-7) [13,23]. Through seven conserved cysteine residues and one conserved histidine residue, the RING finger coordinates two Zn 2+ atoms which stabilize the RING structure [24,25]. The RING finger forms a globular structure with a core three strand β-sheet and a central helix, while the flanking helices align perpendicular to the RING finger ( Figure 2). The RING finger, which is a highly conserved domain found in a large number of proteins, is responsible for the E3-ubiquitin ligase activity of BRCA1 [26]. The N and C-terminal helices are responsible for the interaction of BRCA1 with BARD1 (BRCA1 Associated RING Domain protein 1), a major BRCA1 binding partner that also contains a RING domain [27]. The ubiquitin ligase activity of BRCA1 is dramatically increased by formation of the BRCA1/BARD1 heterodimer [28]. As with all E3-ubiquitin ligases, ubiquitination of a substrate can only occur through interaction with an E2 ubiquitin-conjugating enzyme. UbcH5, as well as other E2 enzymes, binds to the surface of BRCA1 opposite the binding interface with BARD1 [29].
The large number of cancer predisposing mutations that affect the interaction of BRCA1/BARD1 or BRCA1/UbcH5 as well as the RING E3 ligase function suggest that the ubiquitin ligase activity of BRCA1 is essential for its tumor suppressor function (but see [30]). We have gained the most information about the structurefunction of the BRCA1 RING domain from the structure of the BRCA1/BARD1 heterodimer. BARD1 also contains a RING domain with sequence and structural homology to BRCA1, including two flanking alpha helices. The N-terminal alpha-helix of BRCA1 aligns in an antiparallel fashion with the C-terminal alpha helix of BARD1. Conversely, the C-terminal alpha-helix of BRCA1 is antiparallel to the N-terminal alpha-helix of BARD1 ( Figure 2) [13]. The four helix bundle creates a large buried hydrophobic region and stabilizes the heterodimer, while interactions between the BRCA1 RING finger and the flanking alpha-helices maintain the orientation of the RING finger with respect to the flanking alpha-helices. The interaction between BRCA1 and BARD1 both increases the ubiquitin ligase activity of BRCA1 and causes the nuclear export sequence (NES), located on the C-terminal helix of the RING domain of both BRCA1 and BARD1, to be buried [13,32,33]. The buried NES in the four helix bundle results in nuclear retention of the two proteins. The four helix bundle contains the majority of the interactions between BRCA1 and BARD1, however a few inter-RING interactions may occur as well [13]. As stated above, the RING finger of BRCA1 consists of a small three strand antiparallel β-sheet and a central helix. Two Zn 2+ atoms stabilize the structure within the RING finger and are coordinated by Zn 2+ binding loops named Site I and Site II. Site I is made up of four cysteine residues, while Site II contains three cysteine residues and one histidine residue. The Zn 2+ binding residues are highly conserved and characteristic of RING fingers found in many other proteins. Additionally, the spacing between the Zn 2+ binding residues is conserved among many RING fingers. Conversely, a central helix is present in some RING fingers, but not all [13].
Ubiquitination of substrates occurs in a three-step process. First, an E1 ubiquitin-activating enzyme activates a ubiquitin (Ub) molecule, which is transferred to an E2 ubiquitin-conjugating enzyme. The E3 brings together the E2 and substrate to complete the ubiquitination process. The human genome encodes ~40 E2 enzymes, which rely on ~1000 E3 ubiquitin ligases for their specificity [34]. RING E3 ubiquitin ligases, including BRCA1, act solely as scaffolds by binding to the E2 via the RING finger domain, while the substrate binds to another domain on the E3. This brings the substrate close enough to the E2 to allow for transfer of Ub from the E2 to the substrate. The presence of the E2 ubiquitinconjugating enzyme, UbcH5, dramatically increases BRCA1/BARD1 ubiquitination activity in vitro [28].
NMR structures of BRCA1/BARD1/UbcH5c show that loops of UbcH5c bind to a groove formed by the two Zn 2+ binding sites and the central helix of the RING finger of BRCA1, and that UbcH5c has no interaction with BARD1 [29]. Several other E2 proteins have been shown to interact with the BRCA1/BARD1 heterodimer in a yeast-two hybrid study [35]. Targets of BRCA1 E3 ligase activity in vivo include estrogen receptor-alpha, progesterone receptor, CtIP, and histone protein H2A with resulting alterations in gene activation, DNA repair, and DNA condensation [36][37][38][39][40].
Mutation of the cysteine residues that coordinate the Zn 2+ atoms have been reported as clinically important, indicating that they result in altered function and an increased risk of cancer. Mutation of residues in Site I result in altered folding of the RING domain [13]. A more complete study of Site II residue mutations found altered structure by mass spectrometry and reduced Zn 2+ binding at Site II [41] . This study reported that BRCA1/BARD1 heterodimerization was not affected by Site II mutations, however a later study by the same group reported that several Site I and Site II mutations caused not only a decrease in ubiquitin ligase activity, but also a decrease in co-immunoprecipitation of BRCA1 and BARD1 [29]. These studies suggest that mutation of Site I and Site II residues may affect BRCA1 ubiquitin ligase activity by either decreasing BRCA1/BARD1 heterodimerization or BRCA1/UbcH5 interaction. Another study has shown that the E3-ubiquitin ligase activity of BRCA1 is inhibited by platinum (Pt)

Exon 11-13
Exons 11-13 cover over 65% of the sequence of BRCA1 and encode two nuclear localization sequences (NLS) and binding sites for several proteins including retinoblastoma protein (RB), cMyc, Rad50 and Rad51 (reviewed in [44]). The amino acids encoded by these exons also contain portions of a coiled-coil domain which mediates interactions with PALB2, as well as a portion of a serine containing domain (SCD) that is phosphorylated by ATM ( Figure  3). No atomic-level structures have been determined for exons 11-13 of BRCA1. Despite the fact that exons 11-13 contain a large percentage of the clinically relevant mutations, very little is known about the structure or function of this region when compared to the RING or BRCT domains [12]. Interestingly, BRCA1 exon 11-13 binding partners are involved in a wide range of cellular pathways. Myc is a transcription factor for a large number of genes. Rad50, Rad51 and PALB2 are involved in DNA repair. RB controls cell cycle progression. The large number of mutations occurring in this region, many with loss of large portions of sequence, suggests that this region is important for the tumor suppressor function of BRCA1.
The phosphoprotein RB is a well-known tumor suppressor that controls growth by regulating progression through the cell cycle [45].
BRCA1 interacts with the hypo-phosphorylated form of RB via BRCA1 exon 11. Specifically, amino acids 304-394 were found to be responsible for binding to the ABC domain of RB [46]. Overexpression of BRCA1 in cells expressing wild type RB causes suppression of cell cycle progression. Deletion of the region of BRCA1 that mediates BRCA1/RB binding inhibits BRCA1dependent suppression of cell cycle progression [46]. This suggests that the exon 11-mediated interaction between BRCA1 and RB causes cell cycle arrest through actions of RB. This finding also indicates that exon 11 is responsible for BRCA1-dependent cell cycle arrest, and this may also be dependent on the BRCA1/RB interaction.
Rad50 and Rad51 are two proteins involved in DNA repair. Rad50 functions in a complex that includes MreII and Nbs1. This complex is involved in both non-homologous end joining (NHEJ) as well as homologous recombination (HR). An interaction between BRCA1 and Rad50, and therefore with the Rad50/MreII/Nbs1 complex, has been established. This interaction requires BRCA1 exon 11 amino acids 341-748 [47].
BRCA1 recruits the Rad50/MreII/Nbs1 complex to sites of DNA double strand breaks to facilitate DNA repair. BRCA1-null mouse embryonic fibroblast cells exhibit decreased levels of NHEJ activity, which suggest that BRCA1 is involved in the NHEJ process through interaction with the Rad50/MreII/Nbs1 complex. Rad51 is a homologue of the yeast protein RecA and binds to ssDNA, facilitating homologous recombination (HR). BRCA1 is associated with Rad51 during both mitotic and meiotic cells via amino acids 758-1064 [48]. BRCA1 association with Rad50 and Rad51 suggests a role for exon 11 in both NHEJ and HR processes of DNA repair.
The transcription factor c-Myc also interacts with BRCA1. Reports have indicated that c-Myc promotes transcription of up to 15% of the genome, making it a major hub for transcriptional activation [49]. BRCA1 has two c-Myc binding sites (known as MB1 and MB2). MB1 is located only in exon 11 (a.a. 433-511) while MB2 is located in exons 8-11 (a.a. 175-303) [50]. In SVD-P5 cells co-transformed with c-Myc/Ras, transfection with BRCA1 significantly decreased the ability of these cells to form transformed foci [50]. This suggests that the transformation activity of c-Myc/Ras is inhibited by BRCA1 expression. Additionally, the transcriptional activity of Myc is decreased by BRCA1 [50]. Thus, suppression of the oncogenic activities of c-Myc may account for some of the tumor suppressor activity of BRCA1.
Exon 11 contains two nuclear localization sequences (NLS). Amino acids 501-507 (NLS1) and 607-614 (NLS2) are both recognized by importin-α machinery to mediate BRCA1 transport from the cytosol to the nucleus. While both sequences are recognized by importin-alpha, NLS1 is the most critical sequence because mutation of this sequence inhibits all interactions between BRCA1 and importin-alpha [51]. Mutation of the NLS sequences results in altered subcellular localization of BRCA1, with a shift toward cytosolic localization. Clearly, mutations of BRCA1 NLSs causing cytosolic expression of BRCA1 would decrease the tumor suppressor activity of BRCA1 due to the loss of BRCA1's DNA repair activity and subsequent increase in unrepaired mutations and chromosomal abnormalities.
A putative coiled-coil domain spanning exons 11-13 in BRCA1 (a.a. 1364-1437) contains the binding site for PALB2. At this site, PALB2 acts as a scaffold to bring together BRCA1 and BRCA2 to form a complex of the three proteins which is involved in HR during DNA repair. Both BRCA1 and PALB2 contain coiled-coil domains that mediate the interaction of the two proteins. Through modeling of the coiled-coil domain of BRCA1 and PALB2, the interaction sites were mapped to the predicted a-face of the PALB2 helix containing Lys14, Leu21, Tyr28, Leu35, and Glu42 and the predicted a and dfaces of BRCA1 [52]. Mutations in the coiled-coil region of BRCA1 led to the discovery of the PALB2 binding site on BRCA1, since mutations reported in this region (Met1400Val, Leu1407Pro, and Met1411Thr) inhibit interaction between BRCA1 and PALB2 [52]. BRCA1 contains a domain called the serine cluster domain (SCD). A portion of the SCD of BRCA1 is located in exons 11-13, and spans from amino acids 1280-1524. The region has a concentrated amount of putative phosphorylation sites, and is phosphorylated by ATM/ATR kinases in vitro and in vivo. ATM and ATR are kinases activated by DNA damage. Phosphorylation of BRCA1 causes recruitment of BRCA1 to sites of double strand breaks. SCDs are common in ATM/ATR targets including multiple DNA damage response proteins [53]. Serines 1189Serines , 1457Serines , 1524Serines , and 1542 can all be phosphorylated in vivo, while additional serines can be phosphorylated in vitro [54]. Mutation of these serine residues are seen clinically, and may affect localization of BRCA1 to sites of DNA damage and DNA damage response function.

BRCT domain
The BRCA1 C-terminal (BRCT) domain was originally identified in BRCA1, but it is also a conserved domain in multiple other proteins (most being involved in DNA damage repair). BRCT domains can occur as a single BRCT domain, as a tandem repeat (as found in BRCA1), multiple repeats, or fusions between two domains (reviewed in [55]). The BRCA1 BRCT domain mediates phosphoprotein interactions between BRCA1 and proteins phosphorylated by ATM and ATR, two kinases activated by DNA damage (reviewed in [56]). BRCT domains are classified into two categories based on their ability to recognize phosphoproteins. Class-I BRCT domains can recognize phosphoserine (pSer) residues, while Class-II BRCT domains can recognize both pSer and phosphothreonine (pThr) residues. The BRCA1 BRCT domain recognizes the sequence pSer-X-X-Phe in its phosphorylated binding partners, and is therefore a Class-I BRCT domain. Binding partners for the BRCA1 BRCT domain include BACH1, CtIP, and CCDC98/abraxas [57][58][59]. While the main function of the BRCA1 BRCT domain is modulating interactions between BRCA1 and phosphoproteins, BRCT domains, including the BRCT domain of BRCA1, can also mediate DNA binding and non-phosphoprotein interactions [60].
Amino acids 1650-1863 of BRCA1 consist of two tandem BRCT repeats connected by a 22 amino acid linker [14]. Each BRCT repeat consists of three α-helices packed around a four strand β-sheet ( Figure 4A). The two BRCT repeats interact in a head-totail fashion through the interaction between α-helix 2 of BRCT1, and α-helices 1 and 3 of BRCT2 through mainly hydrophobic residues. The architecture of the tandem BRCT allows the BRCA1 BRCT to recognize both a pSer and the 3+ aromatic residue in a bipartite manner in two separate recognition pockets in the cleft between BRCT1 and BRCT2 ( Figure 4B) [15,19] (also reviewed in [55]). The pSer residue forms hydrogen bonds with Ser1655 and Lys1702 and the backbone amine group of Gly1656, all within the N-terminal BRCT1 [19].
The 3+ phenylalanine residue fits into the hydrophobic core created by the two BRCT repeats, while the main chain backbone of the 3+ phenylalanine forms hydrogen bonds with R1699 of α-helix 1 of the N-terminal BRCT domain ( Figure 4C) [19]. The size and subsequent rigidity of the hydrophobic core of the interface between the two BRCT repeats dictates the strict consensus sequence required for substrate recognition by the BRCA1 BRCT domain.
The consensus sequence pSer-X-X-Phe facilitates recognition of targets such as CtIP, BACH1, and abraxas which are all phosphorylated in response to DNA damage (reviewed in [55]).
The BRCA1 BRCT domain has also been shown to bind directly to DNA double strand breaks (DSB) by electron microscopy [60]. However, the in vivo relevance of this interaction is unknown.
While structural studies of the interaction between the BRCA1 BRCT domain and double strand breaks have yet to be carried out, models have been developed with predicted interactions between the BRCT of replication factor RFCp140 and DNA double strand breaks. The BRCT domain of RFCp140 recognizes the terminal 5' phosphate of a 3' overhanging DNA double strand break, as well as the major groove of the DNA adjacent to the double strand break [61]. It is unknown whether or not the BRCT domain of BRCA1 binds to DNA in a similar manner. BRCT domains have also been shown to interact with some proteins in a phosphorylationindependent manner, however this has been much less studied and not well characterized in BRCA1 (reviewed in [55]).
Multiple studies have found mutations in the BRCT domain of BRCA1 in breast and/or ovarian cancers [63][64][65]. Specifically, mutation of hydrophobic residues within the hydrophobic core of the BRCT domain inhibits the ability of BRCA1 to recognize phospholigands [19]. This would suggest that mutation of a residue required for recognition of a substrate would impede the ability of BRCA1 to carry out its role in the DNA damage repair pathway. An interesting example causes BRCA1 to fall into a "similarity trap" [66]. Typically, phosphorylated p53 has a much higher binding affinity for 53BP1 (p53 binding protein 1), than BRCA1. Both 53BP1 and BRCA1 interact with p53 through their tandem BRCT domains, however with different affinities. Two cancer causing mutations in BRCA1, Phe1695Leu and Asp1733Gly cause BRCA1 to bind p53 with similar affinity to 53BP1 [66]. This suggests that these mutations of BRCA1 in the BRCT domain could force BRCA1 into a similarity trap, causing 1) BRCA1 to bind p53 with higher affinity than wild-type BRCA1, and 2) competition for 53BP1 binding to p53. Thus, it is likely that these mutations in the BRCA1 BRCT domain lead to altered p53 function possibly contributing to the cancer phenotype. Another study has shown that cancer causing mutations in other areas of the BRCA1 BRCT domain can alter the backbone structure of the BACH1 binding pocket [67]. This suggests that mutations that affect the BACH1 binding pocket are not limited to just the residues in direct contact with the phosphopeptide. The number of cancer causing mutations in this region suggests that this domain is critical for tumor suppression. BRCT1 and BRCT2 pack together in a head-to-tail orientation and are connected by a linker helix. Helix 2 from BRCT1 and helices 1 and 3 from BRCT2 form a hydrophobic core and stabilize the structure. Rendering was generated using POLYVIEW-3D [31]. Structural model is derived from PDB accession number 1T29. B) The cleft between BRCT1 and BRCT2 forms the binding pocket for proteins phosphorylated by ATM and ATR. The BRCA1 BRCT domains are shown in blue, and a fragment of BACH1 is shown in yellow. C) A magnification of the BRCA1 BRCT/BACH1 binding pocket. The consensus sequence for BRCA1 BRCT recognition of phosphoproteins is 990 pSer-X-X-Phe 993 . The BRCA1-binding region of a phosphopeptide derived from BACH1 is shown. Phospho-Ser990 (pS990) interacts with Ser1655 and Lys1702 of BRCA1, which form a basic pocket. The 3+ Phe993 fits into a hydrophobic pocket created by the two BRCT domains (Phe1704, Met1775, Leu1839). Lysine 995 (K995) forms a salt bridge with Asp1840 and Glu1836. Rendering of B and C was generated using Visual Molecular Dynamics (VMD) [62]. Structural model in B and C is derived from PDB accession number 1T15.